High-Stakes AI
AI applications whose failures could cause significant harm (e.g., medical diagnosis, autonomous vehicles), requiring heightened governance and oversight.
Definition
Systems where errors can lead to physical injury, financial ruin, or societal harm. High-stakes AI demands complete governance rigour: impact and risk assessments, formal verification or clinical trials, human-in-the-loop controls, and real-time monitoring. Certification and regulatory approval (FDA, FAA, EU AI Act) are typically required before deployment, and ongoing compliance audits are mandated.
Real-World Example
An autonomous-drone inspection service for power lines is classified as high-stakes. Governance protocols require flight-simulator testing, formal safety-case documentation, mandatory human-operator remote-override controls, and quarterly safety audits before authorizing any live inspection missions.
.svg)
